JULY 1.5, 1940
ANALYTICAL EDITION
ml. of water. The currentwas then discontinued, and the cathode Tvas n-ithdrawn, rinsed in alcohol, and dried 5 minutes at 110" C. Two copper solutions were prepared by dissolving weighed amounts of Kahlbaum's reagent copper in nitric acid followed by evaporation with sulfuric acid to fumes, diluting, and making up to 250 ml. Five-milliliter aliquot samples were taken for analysis. I t was early determined that 5 or 10 minutes at 2.5 volts seemed sufficient to deposit all the copper. However, occasional erratic results led t o increasing this time to 20 or 30 minutes. Voltages higher than 2.8 volts led to high results and oxidized deposits. Best results were obtained when the voltage was held at about 2.5 volts during most of the electrolysis and then raised t o 3 to 3.5 volts for the remaining 5 minutes. A drop of alcohol amuggested by Uenedetti-Pichler prevented spraying by the emerging gas bubbles. Table I11 gives the results of the determinations of copper in the aliquot solutions using the Pregl and Clarke electrodes and apparatus.
Discussion
It is apparent that the precision with the Clarke apparatus is somewhat better than with the Pregl type. For the Clarke apparatus it is about equal to the precision of the sum of all the other operations except deposition of the metal. This leads to the conclusion that copper can be deposited with a precision better than that of the rest of the electrolytic operation-i. e., that the error in depositing copper is insignificant in comparison with the sum of all the other errors in the complete determination. The mean weight of copper obtained in the second set of results with the Pregl apparatus is considerably lover than the mean of the results with the Clarke apparatus. This supports the general experience in this laboratory when using the Pregl apparatus. It is difficult to avoid touching the cathode when withdrawing it from the cell. This would doubtless scrape off some of the plate and give low results. Further, breaking the circuit while the electrolyte is still in contact wibh the solution is likely to lead to some redissolving of the copper deposit and thus account for the low results obtained with the Pregl apparatus.
433
Summary The magnitudes of the precision of the various operations of a complete microelectrolysis have been estimated in terms of the average deviations of series of observat,ions. Errors in using the microbalance predominate over the errors of handling cathode and n-eights, of cleaning and drying the cathode, and of sampling by aliquots. The precision with which copper can be deposited is somewhat better with the Clarke apparatus than with the Pregl type. For the Clarke apparat'us 1 to 2 mg. of copper can be determined with a precision of 10.010 to *0.012 mg. for a series of determinations. The precision for the complete determination using the Clarke apparat'us (*0.010 mg.) is of the same order of magnitude as for the sum of all the integral operations (*0.010 mg.) except the deposition of the metal. This leads to the conclusion that the errors in depositing copper are insignificant by comparison with the sum of all the other errors.
Literature Cited (1) Benedetti-Pichler, A. A,, ISD. E:sG. C H m f . , Anal. Ed., 8, 373 (1936). (2) Benedetti-Pichler, A. A , , 2. anal. Chem., 62, 321 (1923). (3) Brill, O., and Evans, C. R., J . Chem. Soc.. 93, 1442 (1908). (4) Clarke, B. L., and Hermance, H . IT.,Bell Sustem Tech. Fzibl., Chem. B, No. 658 (1931); J . Am. Chem. S o c . , 54,877 (1932). (5) Emich and Donau, Molonatsh., 30, 745 (1909). ( 6 ) Fales, H. A , , "Inorganic Quantitative .hialysis", p. 61, Nexv York. Aadeton-Centurv Co.. 1939. ( 7 ) Hernler, F.:and Pfenigbkrger, R., Mikrochem., Xolisch Festschrift, 1936, 218. (8) Okacs, -4., Z . anal. Chem., 88, 10s (1932). (9) Phillippi, E . , and Hernler, F., .Ilikrochem., Emich Festschrift, 1930, 241. (10) Pregl, F., "Quantitative organixhe hIikroanalyse", 3rd ed.. D. 185. Berlin. Julius Sarinner. 1930. (11) Rfesenfeld, E. H . , and $Ioll&, H. F., 2. Elektrochcm., 21, 137 (1915). P R E ~ E S T E Dbefore t h e Division
of Microchemistry a t t h e 98th Xeeting of t h e
American Chemical Society, Boston, l l a s s .
Detection of Aromatics in Air G. R. GILBERT AYD R. E. TANNICH, Humble Oil & Refining Company, Baytown, Texas
T
HE processing of aromatic materials in industry involves
a potential hazard to the health of the personnel-exposure to aromatic vapors in the atmosphere. It has been reported ( I ) that 100 p. p. m. of aromatics is the maximum aromatic concentration in which a man may work safely. I n the past, the lack of a method for the determination of aromatics in the atmosphere in the presence of other hydrocarbons made it impractical to define the areas where the concentration of aromatics necessitates protective devices for the workers, in industrial processes such as the sulfur dioxide extraction of petroleum distillates. An analytical procedure that possesses simplicity and a sufficient degree of accuracy has been devised for indicating whether the aromatic concentration presents such a health hazard. The Bureau of Mines ( 2 ) lists most of the methods, based on the measurement of physical properties, reported in the literature-namely, fractional distillation, specific gravity, sorption on solid sorbents, and the use of the interferometer and the spectrograph. Because of the very small quantities of aromatics present, distillation, specific gravity measurements, and gain in weight of a sorbent are inapplicable. The use of the interferometer or the spectrograph requires a skilled operator and expensive equipment.
I n addition to the nitration method described by the Bureau of Mines ( 2 ) , several other methods were investigated. The chemical methods, utilizing the color change brought about by sulfonation or nitration, were found to be unsuited to refinery practice, probably because of organic dust particles in the atmosphere. Attempts to utilize the differences in solubility of picric acid in aromatics and paraffins try titration with caustic or by color differences failed because only traces of aromatics were present in the atmosphere, and the results were inaccurate. Utilization of the difference in index of refraction by absorption in a low-boiling paraffin (n-heptane) was unsatisfactory because at the lorn temperature necessary to prevent "weathering", ice from moist'ure in the atmosphere rapidly plugged the equipment. On the other hand, the use of a high-boiling paraffin resulted in failure because its index of refraction was too close to that of the aromatics.
Description of Apparatus The sample of air to be tested is obtained in a 5-liter Pyrex glass short-necked flask, fitted ivith a tight-fitting rubber stopper through which are inserted two 3- to 4mm. straight-bore glass stopcocks and a low-temperature thermometer (Figure 1). A long, fine capillary tube is used to withdraw the condensed sample from the flask. Indices of refraction are determined with an
VOL. 12, xo. 7
INDUSTRIAL AND ENGINEERIKG CHEMISTRY
434
Calculations TABLE I. D A T .O~N AROXLTICCOXTAVISAXT 1 0 % off on Engler distillation, ' F. 50% off on Engler distillation, F. 90y0 off on Engler distillation, ' F. 243 282 C 298 , M e a n arerage boiling point, 1
+
I n order to calculate the parts of aromatic vapors per 1,000,000 parts of air from the cubic centimeters of hydrocarbons present in 5 liters of the air sample, the following equations are used :
243 262
29s
' F.
Temperature of air I n p u m p r o o m , O F. Absolute pressure of air in p u m p room, mm. Aromatic content of p u m p liquid, fraction b y volume Density of p u m p liquid a t 60' F. Before extraction ~ v i t h4 volumes of 90.5 per cent snlfuric acid After extraction with 4 volumes of 99.3 per cent sulfuric acid Condensation a n d abs-irution of unknown sample from pumD room Refractive index of condrnaate at 6s' F. Condensation and absorption of air containing linown conccntrations of p u m p liquid P u m p liquid, cc. pcr 5 liters of air 0 0005 Aromatics in air, p . p. 111. 19 Refractive indices of condensed liquid a t 08' F. 1 3303 1 3308 1 330.5 1 ,3306
2GS
-_
110 ,JO 0 S>
-
-
('
=
d,
=
density of aromatics at
TO
15.56' C. (GO" F.) present in liquid sample ta!ten from source of contamination
0 8193 0.7500 1.3315
0 0025 03
1,3318 1.3319 1.3320 1.3310
0 0123 k64
1 :i:iGO 1 :1300 1 3816 1.3360
Abbe refractometer at some constant, convenient temperature (either 20' C., 68" F., or tap-water temperature).
Procedure For clarity, the procedure and model calculations are presented in detail. Several 5-Mer flasks are evacuated to about 2 mm. of mercury absolute pressure, and the air sample is obtained by openjng one of the stopcocks in the atmosphere suspected of contamination; two or three air samples should be taken for the purpose of checking. A sample of the liquid causing the contamination is also obtained and the temperature of the air a t the point of sampling . is read and recorded. A portion of the liquid sample is tested for Engler distillation, aromatic content, and density before and after extraction with 4 volumes of 99.5 per cent sulfuric acid. Exactly 0.5 cc. of methyl alcohol is measured into the flask containing the air sample through one of the stopcocks and the alcohol is permitted to evaporate completely in the closed flask by warming up to about 37.78" C. (100' F.). The flask and cont,ents are then cooled to -56.67" C. (-70" F.). A portion of the condensed liquid is removed from the bottom of the flask with the capillary tube and its index of refraction is read at the chosen temperature. Next, 100-cc. blends, containing 0.1, 0.05, 1.25, and 2.5 cc. of liquid contaminant in methyl alcohol, are made up t o obtain the data for a curve of "refractive index of the condensed liquid us. liquid volume in cc.", which is drawn to compare the knoivn concentrations of the hydrocarbon with that present in the unknown air sample. Exactly 0.5 cc. of each of these blends is measured in Thermo m e fer turn into partially evacuated 5-liter flasks and the vacuum is broken by admitting hydrocarbon-free air. The flask is then closed, the liquid is allowed to evaporate and then chilled down according to the procedure outlined in the foregoing paragraph, and the refractive index of the condensed liquid is determined at the chosen temperature. Indices of refraction are plotted against the known volumes of contaminant contained in each 0.5 cc. of the blends to produce a curve from which the cubic centimeters of liquid contained in the unknoxn air sample map be determined. FIGURE 1. h P . 4 R K r U S
(1)
where 5 , represents the aromatic content of the liquid sample in fraction by volume, and d and d, represent, respectively, the density a t 60" F. of the liquid sample before and after extraction with 4 volumes of 99.5 per cent sulfuric acid. 9120
x
cc
x
.Tada
____ X (460 In
+ t) x
760 --
P
=
volumes of aro-
matics yapor per million volunw of air
(2)
n-here cc represents the amount of total hydrocarbons expressed as volume of liquid (in cc.) present in the air sample (as read from the calibration curve of refractive index of condensate tis. cc. of liquid hydrocarbon introduced into the flask), nz the molecular weight of the aromatics, and t and p the temperature in O F. and the absolute pressure in millimeters of mercury, respectively, of the contaminated air. I 338 I
I
I
I
I
1
5 1.337 02
co I-
< 1.335
X W
' n
1.333
W
L I-
o Q 1.331
E W
A A R O M A T I C DIST.
LL
0
002
,004
,006
008
,010
,012
GC. LIQUID IN 5 LITERS AIR
FIGURE 2. REFRACTIVE IXDEX OF COSDESSATE us. CONCENTRATION IN
AIR
Table I shows the data obtained on a sample of aromatic distillate which was contaminating the atmosphere. The curve in Figure 2 has been drawn to fit the data obtained on the 5-liter samples of air containing various known amounts of pump liquid. As shown by the dotted lines, the volume of pump liquid (0.0019 cc.) present in 5 liters of the unknown air sample was determined from the refractive index (1.3315) of the condensate produced. According to the dotted lines of Figure 3, the average molecular weight (112) of the pump liquid is estimated from the mean average boiling point (131.11' C., 268" F.). 0.8493 - (0.15) x 0.7500 Substituting in Equation 1, 0.85
0.867, the specific gravity of the aromatics present in the liquid sample taken from the source of contamination. 0.0019 X 0.83 X 0.867 Substituting in Equation 2 , 9120 X X 112 760 (460 110) X T ~ O= 65 parts per million (by volume) of aromatics present in the unknoTm air sample.
+
Development and Accuracy of 11ethod I n the development of a procedure for determining aromatics in the atmosphere, consideration was given to the fact
JULY 15, 1940 TABLE 11.
ANALYTICAL EDITION DETERMINATION O F
Liquid toluene, cc. per 5 liters of air P. p. m. in air Refractive indices of condensed liquid at 68' F.
TOLUEXE
0.0005 0 . 0 0 2 5
0.005
0.010
0.0125
23
116
232
465
580
1.3332 1.3330 1.3330 1.3330
1.3339 1.3352 1.3340 ,. 1,3340 .. 1.3338 ..
1.3370
1.3375 1,3370 1.3375
.. ..
..
that, if all the hydrocarbons present in the air sample were absorbed in a given volume of a liquid absorbent, it would be possible to calculate the amount of hydrocarbons present in the air from refractive indices of the alcohol-hydrocarbon mixture, the alcohol, and the hydrocarbons, provided a liquid sample of the hydrocarbons present in the air were available.
433
plete vaporization of the liquids to -56.67' C. (-70" F.). It would therefore be impossible to calculate the hydrocarbon vapor present in a sample of air from the qumtity of methyl alcohol injected and refractive indices of the condensate obtained at -70" F. and the methyl alcohol employed. I n order to estimate the concentration of hydrocarbon in t'he unknown air samples, every analysis must therefore be accompanied b y cooling of air samples containing known concentrations of the hydrocarbon distillates in question and the same volume of methyl alcohol as that used in the unknown sample and then const,ructing a curve of "refractive index us. the volume of hydrocarbon distillate used". The amount of hydrocarbon present in the unknoTm sample can then be read from t'his curve by means of the refractive irides of the condensate obtained at -70" F. and the concentration of aromatics expressed in terms of parts per million can be calculated. An inspection of the distribution of the points which determined the curves for both toluene and the aromatic distillate in Figure 2 shows that, for concentrations in the range of 100 p. p. m. (0.00215 cc. of toluene and 0.0027 cc. of aromatic distillate per 5 liters of air), the method is accurate to about 15 p. p. m. of aromatics. For concentrations in the range of 400 p. p. m. (0.0086 cc. of toluene and 0.0108 cc. of aromatic distillate per 5 liters of air), the method is accurate to only about 80 p. p. m.
Derivation of Equation for Determining Aromatic Vapors i n Air Let cc
=
d = d, = d, =
za = m =
100 200 300 400 MOLECULAR WEIGHT
WEIGHTOF HYDROFIGURE 3. MOLECULAR CARBONS US. MEANBOILING POIST Methyl alcohol was chosen as the absorbent because its refractive index is less than that of the aromatic hydrocarbons and is about the same as that of water. The former relationship makes possible the calculations required, and the latter relationship minimizes the error introduced by variations in humidity of the air sample. I n testing out the possibilities of applying the foregoing principles to determine the amount of aromatics present in the atmosphere, blends of 0.0005, 0.0025, 0.0050, 0.010, and 0.0125 cc. of liquid toluene with 0.5 cc. of methyl alcohol were measured into 5-liter flasks at 26.67" C. (80" F.) and a n absolute pressure of 760 mm., evaporated in the closed flasks by warming the mixture to about 37.78" C. (100" F.), and then cooled to -56.67" C. (-70" F.). Table I1 shows refractive indices of the condensate obtained in each case. These data, which are plotted in Figure 2 , show that, since the refractive index of the condensate was not equal to that of a mixture of the methyl alcohol and hydrocarbons injected into the flask, only a partial condensation of both the methyl alcohol and the hydrocarbon vapors must have been obtained when the vapors were cooled from the temperature of com-
t
=
p
=
cubic centimeters of liquid hydrocarbons contained in 5 liters of the contaminated air density a t 60" F. of liquid hydrocarbons density a t 60" F. of liquid hydrocarbons after extractionwith 4 volumes of 99.5 per cent sulfuric acid density a t 60" F. of liquid aromatics removed by extraction with 4 volumes of 99.5 per cent sulfuric acid volumetric fraction of aromatics in hydrocarbon sample (determined by extraction with 4 volumes of 99.5 per cent sulfuric acid) average molecular weight of liquid aromatics temperature of contaminated air at time of sampling, F. absolute pressure of contaminated air at time of sampling, millimeters of mercury
- (' -
specific gravity of liquid aromatics cc X rado = grams of aromatic vapor in 5 liters of air at t ' F. and absolute pressure, p mm. of Hg cc X zoda ~- moles of aromatic vapor in 5 liters of air at t F. m and absolute pressure, p mm. of Hg cc ~- ' a d o - moles of aromatic vapor in 1 liter of air a t t 0 F. 5m and absolute uressure. P mm. of Hacc X rodo 460 t 760 460 32 X - = moles of aromatic vapor in 1 tjm liier of air under standard conditions of temperature and pressure = d, =
Xa
+
I
.
+
~
+
460 t 760 X- 460 32 x - x 1,000,000 = moles of aromatic 5m P vaDor in 1.000.000 liters of air under standard conditions of , , temperature and pressure
cc X xoda
+
cc
'
'oda 460 + X 760 X 1,000,000 X 22.4 = liters of aro5m x - 492 P matic vapor in 1,000,000 liters (parts per million) of air under standard conditions of temperature and pressure
Collecting constants except that for pressure, 22 4 x 1,000,000 cc x z4d. (460 + t ) x __ 760 = . x m 0 -i ,. 492 -_760 9120 x x (460 + t ) x = p. p. m. of aromatics
e n rn
P
KO attempt is made to correct for the relative volatilities of the components.
INDUSTRIAL AND ENGINEERING CHEMISTRY
436
Summary The analytical method devised for estimating the atmospheric aromatic contents consists of cooling a blend of the unknown air sample and the vapor of a fixed volume of methyl alcohol on one hand, and blends of air samples containing known concentrations of the aromatic distillates in question and the vapors of a fixed volume of methyl alcohol on the other, to -56.67" C. (-70" F.) and obtaining refractive indices of the condensates. This method is reasonably simple and gives a n accuracy
VOL. 12, NO. 7
within 15 parts per million for critical concentrations of 100 parts per million.
Acknowledgment The authors extend thanks to the Humble Oil and Refining Company for permission to publish their findings.
Literature Cited National Safety Council, 1926. (2) Yant, IT.P., Pearce, S. J., and Schrenk,H. H., U.S.Bur. Mines, R e p t . Inwstigations 3323, 1936. (1) Benzol Report,
Microscope Hot Stage for Determination of Melting Points Application to Carotenoid Pigments F. P. ZSCEIEILE
AND
J. W. WHITE, JR., Purdue University, Lafayette, Ind.
T
HE usual method of determining melting points by means
of aBerl block or a liquid bath surrounding a melting point tube does not give satisfactory results when applied to such compounds as the carotenoid pigments. The melting point of these compounds observed in this manner is really an indication of decomposition; it is not a sharply defined point and considerable material is required. Values for the melting point of alpha-carotene reported in the literature vary from 178.5" (11) to 188" C. ( 8 ) , a range of 9.5" C. For betacarotene the reported values vary from 177.8" (11) to 187.5"C. (IO),a range of 9.7" C. The range of any one determination is usually 1" C. or more. The sintering point is usually observed slightly below the melting range. A number of hot stages for melting point determination hare been described ( 1 , 3 , 4 , 5 , 9 ) ,but the apparatus discussed here includes refinements which add greatly to the precision of measurement. T o the authors' knowledge, this technique has not been applied previously to carotenoid pigments, to which i t is very well adapted. I t s use should add significance to the melting point as a physical constant in carotenoid chemist'rg.
Description of Apparatus The hot stage is constructed as shown in Figure 1. The two copper plates are placed on the stage of a polarizing microscope, and the sample is placed on a circular cover glass in the upper recess of the lorver plate. A4spring clip holds the cover glass in place, so that the crystals are directly over the 0.156-cm. (0.0625inch) hole passing through the axis of the copper plates. The sample is thus illuminated from below Iyith polarized light and is observed through the analyzing S i c o l prism of the microscope. The stage is heated electrically by means of a coil of 60 cm. ( 2 feet) of S o . 21 Xichrome wire having a resistance of 3.1 ohms. The core, machined from lava and baked to hardness, is recessed in the bottom of the lower block around the central hole. An annulus of lava is cemented under the heating element to enclose it within the block. Control of the heating rate is effected by a Variac variable transformer, Type 200-C. Heating rates of from 0.1" to 6.5" C. per minute can be maintained at any temperature from that of the room t o 250' C. The temperature within the apparatus is measured by a Leeds & Xorthrup portable precision potentiometer and an iron-constantan thermocouple. The hot junction of the couple, about 1 mm. in diameter, is located within 1 or 2 mm. of the sample. The leads are brought diagonally through the lower plate in a porcelain tube, cemented in place. Two holes lead from the outside to the central chamber for the passage of nitrogen through the apparatus. Figure 2 shom how the hot stage is insulated from the microscope stage by a Pyrex support on an asbestos plate. A Leitz
Periplan 20-power ocular and a Bausch & Lomb 4-power objective with a 38-mm. working distance provide suitable magnification and permit the objective t o be sufficiently distant from the hot stage t o avoid heat damage. A camera lucida makes it possible to observe simultaneously both the sample being melted and the galvanometer pointer. At the left center of the photograph is the Variac transformer, with voltmeter. To the right of the microscope is the potentiometer box; at the extreme right is a Dewar flask containing the cold junction of the thermocouple. In front of the Variac is a bubble counter to measure the flow- of nitrogen. The precision pinchcock in the foreground regulates the rate of gas flow. A micro combustion tube containing heated copper for removal of traces of oxygen from the nitrogen and a drying tube containing Dehydrite are not shown. With this equipment, the thermocouple voltage can be estimated to 1 microvolt, which corresponds to 0.02" C. with an iron-constantan thermocouple.
Calibration of Apparatus A heating rate of 0.3" C. per minute was selected because
it is slow enough to permit the sample and the thermocouple to attain the same temperature and i t is rapid enough to avoid an unduly long time for the determination. The thermocouple voltage may be kept balanced easily a t this heating rate. The voltage necessary to maintain a heating rate of 0.3" C. per minute at any desired temperature was determined by measurement of the block temperature increase with time for four voltages. Rates of heating were plotted against temperature for each voltage. The temperature a t which a given voltage heated the apparatus at the desired rate (0.3" C. per minute) was then plotted against that voltage. The result was a straight line from which can be read the voltage necessary to heat the block a t 0.3" C. per minute at any temperature from room temperature to 250" C. When a melting point is to be determined, the sample, which may be a single crystal if necessary, is placed in position and the apparatus is heated at 22.5 volts, corresponding to a current o,f 6 amperes, the upper limit of the heating element. Within 1.5 to 2.0° of the anticipated melting point, the voltage is set at the point necessary to maintain a heating rate of 0.3" per minute. The response to voltage change is sufficiently rapid t o allow maintaining solid and liquid phases as desired. Thus, if the sample does not decompose, several melting points may be determined with the same sample. Nitrogen should flow through the apparatus during the entire heating procedure. About 1.5 minutes before the expected melting point is reached, the flow of nitrogen is stopped to ensure temperature equilibrium. When anisotropic crystals are observed between crossed Kicol prisms, the melting point is observed as a sudden disappearance of
